SpatialBench evaluates 41 spatial foundation models across 6 paradigms and 5 task suites, finds they are not all-round players, and introduces the DA-Next-5M dataset plus DA-Next baseline model.
Depth anything with any prior
13 Pith papers cite this work. Polarity classification is still indexing.
13
Pith papers citing it
abstract
This work presents Prior Depth Anything, a framework that combines incomplete but precise metric information in depth measurement with relative but complete geometric structures in depth prediction, generating accurate, dense, and detailed metric depth maps for any scene. To this end, we design a coarse-to-fine pipeline to progressively integrate the two complementary depth sources. First, we introduce pixel-level metric alignment and distance-aware weighting to pre-fill diverse metric priors by explicitly using depth prediction. It effectively narrows the domain gap between prior patterns, enhancing generalization across varying scenarios. Second, we develop a conditioned monocular depth estimation (MDE) model to refine the inherent noise of depth priors. By conditioning on the normalized pre-filled prior and prediction, the model further implicitly merges the two complementary depth sources. Our model showcases impressive zero-shot generalization across depth completion, super-resolution, and inpainting over 7 real-world datasets, matching or even surpassing previous task-specific methods. More importantly, it performs well on challenging, unseen mixed priors and enables test-time improvements by switching prediction models, providing a flexible accuracy-efficiency trade-off while evolving with advancements in MDE models.
citation-role summary
method 2
background 1
citation-polarity summary
representative citing papers
A boundary-forcing masked modeling paradigm for self-supervised vision pretraining yields a 1B model rivaling 7B models on dense spatial perception tasks.
SLIM adapts MoGe-2 to truly sparse LiDAR via partial-convolution encoder and multi-scale fusion neck, cutting absolute relative depth error by 39-51% at 100-150 m on Virtual KITTI and CARLA under density-agnostic training.
Cross3R performs feed-forward 3D reconstruction and 6-DoF pose estimation from any combination of satellite, UAV, and ground images, outperforming baselines on a new 278K-image tri-view dataset.
An automated real-to-sim pipeline builds digital twins and affordance-preserving cousins from video, yielding sim evaluations that correlate with real robot policy success and zero-shot sim-to-real gains.
Presents the WildCross benchmark with 476K frames for place recognition and metric depth estimation in natural environments, demonstrating limitations of existing vision models.
DeblurNVS restores geometric representations via latent diffusion to enable high-fidelity novel view synthesis directly from sparse motion-blurred inputs.
FocusDepth is a prompt-conditioned framework that fuses SAM3 features into Depth Anything models via Multi-Scale Spatial-Aligned Fusion to improve target-region depth accuracy on the new FDE-Bench.
Unlabeled web videos processed by designed data engines generate effective training data that yields strong zero-shot and finetuned performance on 3D detection, segmentation, VQA, and navigation.
FoundDP integrates DP-derived metric depth with ViT-based structural priors from monocular models, using feature alignment to mitigate defocus blur and improve depth in low-observability areas.
Local RANSAC alignment of Depth Anything-style priors to raw sensor depth recovers metric glass-surface geometry without training, and the new GlassRecon benchmark shows large gains on severely corrupted hard samples.
ViPE estimates camera intrinsics, motion, and dense near-metric depth from uncalibrated videos, outperforming baselines on TUM and KITTI while releasing annotations for 96M frames across real and generated videos.
LDCM achieves state-of-the-art metric depth completion from sparse observations by combining foundation-model initialization with a point-map regression head that removes the need for camera intrinsics.
citing papers explorer
-
SpatialBench: Is Your Spatial Foundation Model an All-Round Player?
SpatialBench evaluates 41 spatial foundation models across 6 paradigms and 5 task suites, finds they are not all-round players, and introduces the DA-Next-5M dataset plus DA-Next baseline model.
-
Vision Pretraining for Dense Spatial Perception
A boundary-forcing masked modeling paradigm for self-supervised vision pretraining yields a 1B model rivaling 7B models on dense spatial perception tasks.
-
Sparse-LiDAR Prompting of Monocular Geometry Foundations: An Empirical Study Toward Long-Range Driving Depth
SLIM adapts MoGe-2 to truly sparse LiDAR via partial-convolution encoder and multi-scale fusion neck, cutting absolute relative depth error by 39-51% at 100-150 m on Virtual KITTI and CARLA under density-agnostic training.
-
Seeing Across Skies and Streets: Feedforward 3D Reconstruction from Satellite, Drone, and Ground Images
Cross3R performs feed-forward 3D reconstruction and 6-DoF pose estimation from any combination of satellite, UAV, and ground images, outperforming baselines on a new 278K-image tri-view dataset.
-
SimFoundry: Modular and Automated Scene Generation for Policy Learning and Evaluation
An automated real-to-sim pipeline builds digital twins and affordance-preserving cousins from video, yielding sim evaluations that correlate with real robot policy success and zero-shot sim-to-real gains.
-
Cross-Modal Benchmarking for Robotic Perception in Natural Environments
Presents the WildCross benchmark with 476K frames for place recognition and metric depth estimation in natural environments, demonstrating limitations of existing vision models.
-
DeblurNVS: Geometric Latent Diffusion for Novel View Synthesis from Sparse Motion-Blurred Images
DeblurNVS restores geometric representations via latent diffusion to enable high-fidelity novel view synthesis directly from sparse motion-blurred inputs.
-
Focusable Monocular Depth Estimation
FocusDepth is a prompt-conditioned framework that fuses SAM3 features into Depth Anything models via Multi-Scale Spatial-Aligned Fusion to improve target-region depth accuracy on the new FDE-Bench.
-
Lifting Unlabeled Internet-level Data for 3D Scene Understanding
Unlabeled web videos processed by designed data engines generate effective training data that yields strong zero-shot and finetuned performance on 3D detection, segmentation, VQA, and navigation.
-
FoundDP: Revisiting Weak Disparity Observability in Dual-Pixel Depth Estimation
FoundDP integrates DP-derived metric depth with ViT-based structural priors from monocular models, using feature alignment to mitigate defocus blur and improve depth in low-observability areas.
-
Enhancing Glass Surface Reconstruction via Depth Prior for Robot Navigation
Local RANSAC alignment of Depth Anything-style priors to raw sensor depth recovers metric glass-surface geometry without training, and the new GlassRecon benchmark shows large gains on severely corrupted hard samples.
-
ViPE: Video Pose Engine for 3D Geometric Perception
ViPE estimates camera intrinsics, motion, and dense near-metric depth from uncalibrated videos, outperforming baselines on TUM and KITTI while releasing annotations for 96M frames across real and generated videos.
-
Large Depth Completion Model from Sparse Observations
LDCM achieves state-of-the-art metric depth completion from sparse observations by combining foundation-model initialization with a point-map regression head that removes the need for camera intrinsics.